feat: allow actor state processor to run without leasing #178
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Leasing of actor states to process causes high contention on the visor_processing_actors table since many instances are competing to lock rows to update their leases. With this change, when --actorchange-lease is set to zero then the actor change processor will not use leasing at all. Instead it assumes that it can work on any actor states between the height ranges specified with --from and --to. It further subdivides that height range by the number of workers started so each worker gets its own range of heights to process. Note that this requires manual coordination of height ranges allocated to each instance of visor otherwise each instance may duplicate work performed by another instance. Note also that --to should be set to a reasonable value, typically under the current height of the chain. Since the range specified with --to and --from is subdivided evenly between workers a very large upper bound could result in some workers being allocated ranges that have no data on the chain.
placer14
reviewed
Oct 30, 2020
| if heightFrom > heightTo { | ||
| return xerrors.Errorf("--from must not be greater than --to") | ||
| } | ||
|
|
willscott
approved these changes
Oct 30, 2020
|
Applies to #126. |
placer14
added a commit
that referenced
this pull request
Nov 4, 2020
By Ian Davis (25) and others Via GitHub origin/master: (66 commits) feat: add changelog generator (#199) feat: disable leasing when gas outputs lease time is zero (#198) feat: add power actor claim extraction chore: Include RC releases in push docker images (#195) fix(metrics): export the completion and batch selection views (#197) fix: fix actor completion query chore: add metrics to leasing and work completion queries continue message parsing task when cbor parse fails (#185) perf: ensure processing updates always include height in criteria (#192) perf: include height restrictions in update clauses of leasing queries (#189) open lotusrepo lens read only feat: allow actor state processor to run without leasing (#178) feat: rpc reconnection on failure (#149) chore: add changelog (#150) polish: update miner processing logic feat: add dynamic panel creation based on tags (#159) don't codeql on every commit to pr. only on PR merge and cron (#168) track statechange persist duration feat: add dynamic panel creation based on tags Replace filecoin-ffi with stubs to simplify build process (#166) ... Conflicts: commands/run.go It looks like you may be committing a merge. If this is not correct, please remove the file /Users/mg/work/pl/sentinel/.git/modules/sentinel-visor/MERGE_HEAD and try again.
bugs-404
reviewed
Aug 4, 2021
| func estimateCurrentEpoch() int64 { | ||
| return int64(time.Since(time.Date(2020, 8, 24, 22, 0, 0, 0, time.UTC)) / (30 * time.Second)) | ||
| return int64(time.Since(mainnetGenesis) / (builtin.EpochDurationSeconds)) |
There was a problem hiding this comment.
func estimateCurrentEpoch() int64 {
return int64(time.Since(mainnetGenesis) / (builtin.EpochDurationSeconds * time.Second))
}
There was a problem hiding this comment.
Can you open an issue if you think there is a problem here
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Leasing of actor states to process causes high contention on the visor_processing_actors
table since many instances are competing to lock rows to update their leases.
With this change, when --actorchange-lease is set to zero then the actor change processor
will not use leasing at all. Instead it assumes that it can work on any actor states
between the height ranges specified with --from and --to. It further subdivides that
height range by the number of workers started so each worker gets its own range of
heights to process.
Note that this requires manual coordination of height ranges allocated to each
instance of visor otherwise each instance may duplicate work performed by another
instance.
Note also that --to should be set to a reasonable value, typically under the current
height of the chain. Since the range specified with --to and --from is subdivided
evenly between workers a very large upper bound could result in some workers being
allocated ranges that have no data on the chain.